This is the s390 port of 70627654 "x86, extable: Switch to relative
exception table entries".
Reduces the size of our exception tables by 50% on 64 bit builds.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The cpu polarization member is the only per cpu state that is not part
of the pcpu structure. So add it there and have everything in one place.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The maximum nesting of the cpu topology is evaluated when /proc/sysinfo
is the first time read. This happens without a lock and a concurrent
reader on a different cpu can see and use an invalid intermediate value.
Besides the fact that this race is quite unlikely the worst thing that
could happen is that /proc/sysinfo would contain bogus information about
the machine's cpu topology.
Nevertheless this should be fixed. So move the detection code to the
early machine detection code and since now the value is early available
use it in the topology code as well.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Add a couple of missing fields that were introduced with z196.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The current proc implementation of the /proc/sysinfo file writes all
informations contained in all system information blocks to a single
page.
This is done by calling sprintf all the time in the expectation that
everything will fit into a single page. This however is not necessarily
true if the configuration of a machine is very large.
So convert /proc/sysinfo to avoid writing into random memory regions.

For readability reasons a couple of lines are longer than 80 characters.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Any change to sysinfo.h causes a whole kernel recompile since sysinfo.h is
included by topology.h, which again is used nearly everywhere.
So remove that include and add a forward declaration instead.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Shorten the code for addressing mode initialization. Also add missing
__init annotations, since this code is only used during kernel initialization.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Renaming the globally visible variable "user_mode" to "addressing_mode" in
order to fix a name clash was not a good idea. (Commit 37fe1d73 "s390/mm:
rename user_mode variable to addressing_mode")
Looking at the code after a couple of weeks one thinks: addressing mode of
what?
So rename the variable again. This time to s390_user_mode. Which hopefully
makes more sense.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Change the default addressing mode so that user space runs in primary space
and the kernel runs in home space.
In addition remove the "switch_amode" kernel parameter so all users who
already specified they want the new default behaviour will stay in the
"switched" mode instead of in the opposite they intended.
If there is a need to switch addressing modes, this can be done with the
"user_mode" kernel parameter: user_mode=home
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

When a sigp instruction is issued it may store a status. This status is
currently stored in a per cpu field of the target cpu.
If multiple cpus issue a sigp instruction with the same target cpu
and a status is stored the result is not necessarily as expected.

Currently this is not an issue:
- on cpu hotplug no sigps, except "restart" and "sense" are sent to the
  target cpu.
- on external call we don't look at the status if it is stored
- on sense running the condition code "status stored" is sufficient to
  tell if a cpu is running or not
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Add a line for each cpu cache to /proc/cpuinfo.
Since we only have information of private cpu caches in sysfs we
add a line for each cpu cache in /proc/cpuinfo which will also
contain information about shared caches.

For a z196 machine /proc/cpuinfo now looks like:

vendor_id       : IBM/S390
bogomips per cpu: 14367.00
features        : esan3 zarch stfle msa ldisp eimm dfp etf3eh highgprs
cache0          : level=1 type=Data scope=Private size=64K line_size=256 associativity=4
cache1          : level=1 type=Instruction scope=Private size=128K line_size=256 associativity=8
cache2          : level=2 type=Unified scope=Private size=1536K line_size=256 associativity=12
cache3          : level=3 type=Unified scope=Shared size=24576K line_size=256 associativity=12
cache4          : level=4 type=Unified scope=Shared size=196608K line_size=256 associativity=24
processor 0: version = FF,  identification = 000123,  machine = 2817
processor 1: version = FF,  identification = 100123,  machine = 2817
processor 2: version = FF,  identification = 200123,  machine = 2817
processor 3: version = FF,  identification = 200123,  machine = 2817
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Allow user-space processes to use transactional execution (TX).
If the TX facility is available user space programs can use
transactions for fine-grained serialization based on the data
objects that are referenced during a transaction. This is
useful for lockless data structures and speculative compiler
optimizations.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Allow user-space threads to use runtime instrumentation (RI). To enable RI
for a thread there is a new s390 specific system call, sys_s390_runtime_instr,
that takes as parameter a realtime signal number. If the RI facility is
available the system call sets up a control block for the calling thread with
the appropriate permissions for the thread to modify the control block.

The user-space thread can then use the store and modify RI instructions to
alter the control block and start/stop the instrumentation via RION/RIOFF.

If the user specified program buffer runs full RI triggers an external
interrupt. The external interrupt is translated to a real-time signal that
is delivered to the thread that enabled RI on that CPU. The number of
the real-time signal is the number specified in the RI system call. So,
user-space can select any available real-time signal number in case the
application itself uses real-time signals for other purposes.

The kernel saves the RI control blocks on task switch only if the running
thread was enabled for RI. Therefore, the performance impact on task switch
should be negligible if RI is not used.

RI is only enabled for user-space mode and is disabled for the supervisor
state.
Reviewed-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Add support for EADM interrupt statistics in /proc/interrupts.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

machine_crash_shutdown() was the only function in crash.c.
So move the function and delete one file.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Unify all our cpu hotplug notifiers to mask out the CPU_TASKS_FROZEN
bit, so we don't have to add all the *_FROZEN variant cases to the
notifiers.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Saves a couple of lines of code.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Expose cpu cache topology via sysfs.
The created sysfs directory structure is compatible to what x86, ia64
and powerpc have.
On s390 we expose only information about cpu caches which are private
to a cpu via sysfs . Caches which are shared between cpus do not have
a sysfs representation.
The reason for that is that the file "shared_cpu_map" is mandatory
and only if running under LPAR it is possible to tell which cpus
share which cache. Second level hypervisors however do not and cannot
expose that information to guests.
In order to have a consistent view we made the choice to always only
expose information about private cpu caches via sysfs.

Example for a z196 cpu (cpu1 in /sys/devices/cpu):

cpu1/cache/index0/size -- 64K
cpu1/cache/index0/type -- Data
cpu1/cache/index0/level -- 1
cpu1/cache/index0/number_of_sets -- 64
cpu1/cache/index0/shared_cpu_map -- 00000000,00000002
cpu1/cache/index0/shared_cpu_list -- 1
cpu1/cache/index0/coherency_line_size -- 256
cpu1/cache/index0/ways_of_associativity -- 4
cpu1/cache/index1/size -- 128K
cpu1/cache/index1/type -- Instruction
cpu1/cache/index1/level -- 1
cpu1/cache/index1/number_of_sets -- 64
cpu1/cache/index1/shared_cpu_map -- 00000000,00000002
cpu1/cache/index1/shared_cpu_list -- 1
cpu1/cache/index1/coherency_line_size -- 256
cpu1/cache/index1/ways_of_associativity -- 8
cpu1/cache/index2/size -- 1536K
cpu1/cache/index2/type -- Unified
cpu1/cache/index2/level -- 2
cpu1/cache/index2/number_of_sets -- 512
cpu1/cache/index2/shared_cpu_map -- 00000000,00000002
cpu1/cache/index2/shared_cpu_list -- 1
cpu1/cache/index2/coherency_line_size -- 256
cpu1/cache/index2/ways_of_associativity -- 12
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

So far, large page support was completely disabled with
CONFIG_DEBUG_PAGEALLOC, although it would be sufficient if only the
large page kernel mapping was disabled. This patch enables large page
support with CONFIG_DEBUG_PAGEALLOC, while it prevents the large kernel
mapping in that case.
Signed-off-by: NGerald Schaefer <gerald.schaefer@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Our memcpy and memcmp variants were implemented by calling the corresponding
gcc builtin variants.
However gcc is free to replace a call to __builtin_memcmp with a call to memcmp
which, when called, will result in an endless recursion within memcmp.
So let's provide asm variants and also fix the variants that are used for
uncompressing the kernel image.
In addition remove all other occurences of builtin function calls.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The s390 implementation of the JIT compiler for packet filter speedup.
Reviewed-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The bit for high gprs in the AT_HWCAP auxiliary vector field and the
highgprs tag in the output of /proc/cpuinfo should not be set for
31 bit kernels.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The native 31 bit and the compat behaviour for the mmap system calls differ:

In native 31 bit mode the passed in address for the mmap system call will be
unmodified passed to sys_mmap_pgoff().
In compat mode however the passed in address will be modified with
compat_ptr() which masks out the most significant bit.

The result is that in native 31 bit mode each mmap request (with MAP_FIXED)
will fail where the most significat bit is set, while in compat mode it
may succeed.

This odd behaviour was introduced with d3815898 "[S390] mmap: add missing
compat_ptr conversion to both mmap compat syscalls".

To restore a consistent behaviour accross native and compat mode this
patch functionally reverts the above mentioned commit.

Cc: stable@vger.kernel.org
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The compat wrappers incorrectly called the non compat versions of
the system process_vm system calls.

Cc: stable@vger.kernel.org
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

There are multiple errors in how sys_32_personality() handles personality
flags stored in top three bytes.

- directly comparing current->personality against PER_LINUX32 doesn't work
  in cases when any of the personality flags stored in the top three bytes
  are used.
- directly forcefully setting personality to PER_LINUX32 or PER_LINUX
  discards any flags stored in the top three bytes

Fix the first one by properly using personality() macro to compare only
PER_MASK bytes.
Fix the second one by setting only the bits that should be set, instead of
overwriting the whole value.
Signed-off-by: NJiri Kosina <jkosina@suse.cz>
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

We use the user_mode() helper already at several places but also
have the open coded variant at other places.
Convert the code to always use the helper function.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Fix name clash with user_mode() define which is also used in common code.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Currently the vmcmd shutdown action is parsed by the kernel and
if multiple cp commands have been specified, they are issued
separately with the cpcmd() function.

The underlying diagnose 8 instruction already allows to specify
multiple commands that are separated by 0x15. The ASCEBC() function
used by cpcmd() translates '\n' to 0x15. The '\n' character is
currently used as vmcmd command separator and therefore the vmcmd
string can be passed directly to the cpcmd() function.

Using the diagnose 8 command separation has the advantage that also
after disruptive commands that stop Linux, for example "def store",
additional commands can be executed.
Signed-off-by: NMichael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Use RO_DATA_SECTION instead of RODATA like several other archs do already.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Follow x86 and MIPS and sort the main exception table at build time.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

debug.c is not a module, so remove the module_exit function, since it
will never be called.
Also move the EXPORT_SYMBOL statements to the functions they belong to.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The current virtual timer interface is inherently per-cpu and hard to
use. The sole user of the interface is appldata which uses it to execute
a function after a specific amount of cputime has been used over all cpus.

Rework the virtual timer interface to hook into the cputime accounting.
This makes the interface independent from the CPU timer interrupts, and
makes the virtual timers global as opposed to per-cpu.
Overall the code is greatly simplified. The downside is that the accuracy
is not as good as the original implementation, but it is still good enough
for appldata.
Reviewed-by: NJan Glauber <jang@linux.vnet.ibm.com>
Reviewed-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Reviewed-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Remove the file name from the comment at top of many files. In most
cases the file name was wrong anyway, so it's rather pointless.

Also unify the IBM copyright statement. We did have a lot of sightly
different statements and wanted to change them one after another
whenever a file gets touched. However that never happened. Instead
people start to take the old/"wrong" statements to use as a template
for new files.
So unify all of them in one go.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>

If lgr has not been initialized, the lgr_info_log() function currently
crashes because 'lgr_page' is not allocated. To fix this 'lgr_page'
is allocated statically now.
Signed-off-by: NMichael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Just saves a couple of instructions.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The s390 idle accounting code uses a sequence counter which gets used
when the per cpu idle statistics get updated and read.

One assumption on read access is that only when the sequence counter is
even and did not change while reading all values the result is valid.
On cpu hotplug however the per cpu data structure gets initialized via
a cpu hotplug notifier on CPU_ONLINE.
CPU_ONLINE however is too late, since the onlined cpu is already running
and might access the per cpu data. Worst case is that the data structure
gets initialized while an idle thread is updating its idle statistics.
This will result in an uneven sequence counter after an update.

As a result user space tools like top, which access /proc/stat in order
to get idle stats, will busy loop waiting for the sequence counter to
become even again, which will never happen until the queried cpu will
update its idle statistics again. And even then the sequence counter
will only have an even value for a couple of cpu cycles.

Fix this by moving the initialization of the per cpu idle statistics
to cpu_init(). I prefer that solution in favor of changing the
notifier to CPU_UP_PREPARE, which would be a different solution to
the problem.

Cc: stable@vger.kernel.org
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The smp and the kvm code have different defines for the sigp order codes.
Let's just have a single place where these are defined.
Also move the sigp condition code and sigp cpu status bits to the new
sigp.h header file.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>